Pure//Accelerate techfest22 Recap
In This Article
Pure Storage's yearly tech conference called Pure//Accelerate occurred June 8th both virtually and in-person in Los Angeles, CA. With the last couple of years being almost exclusively virtual for most OEM conferences, Pure chose to start small for the in-person portion this year by limiting the conference to partners and a small number of customers. All the in-person sessions and keynotes were broadcasted live and made available on the event website, along with many more breakout sessions available on-demand that were not available at the in-person event. I do have to say, it was nice to be back in person to see some familiar faces along with making some new connections. I look forward to next year's conference being back in full swing and being open to all of Pure and WWT's customers.
The one-day conference kicked off with a Keynote lead by Pure's CEO – Charlie Giancarlo and closed with an appearance by Pure's Founder - John 'Coz' Colgrove. The theme for this year's event was "Uncomplicate Data Storage, Forever."
To prove Pure's commitment to keeping all of their products uncomplicated, they showed a video from 2012 where Coz's 7-year-old son setup the companies first FlashArray product by following a simple set of instructions to cable up the disk shelves to the stand alone controllers (think FA-320 days) and configure the Volumes and Hosts. The video then fast-forwarded to 2022 showing a young girl deploying a DB cluster via Portworx Data Service's simple to use, single-click deployment user interface.
This year's conference was focused around the following 3 key areas:
- Cloud operating model
- Cloud native applications
- All-flash data center
Almost a decade ago, Pure was revolutionary in releasing the concept of Evergreen Storage. Evergreen Storage is not just a business model, but an engineering philosophy. With an Evergreen Gold subscription service, your storage never gets old through the following Evergreen Gold advantages:
- Ever Modern (previously known as "Free Every Three") – The storage controllers are replaced with a new model every three years to keep your storage current thus removing the need to migrate data via planned downtime, outages, or performance hits.
- Ever Agile (previously known as "Anytime Controller Upgrades") – With Pure1's ability to trend capacity usage and plan for growth, what happens when your workloads change, and you need larger storage arrays? Don't worry, simply upgrade to larger controllers (while getting a trade-in credit for the investment you've already made) and easily swap out your current controllers non-disruptively for larger controllers with increased performance without needing to migrate data.
- Capacity Consolidation – As storage modules become more and more dense, keep your storage array footprint small in the data center by taking advantage of trading in your smaller flash storage modules for a credit and purchase larger (denser) capacity DirectFlash Modules. Then simply do non-disruptive internal data migrations on the array. In the end, you'll keep your data center power and cooling costs low while keeping up with Pure's philosophy of never buying the same storage capacity again.
That all sounds great, but what was released this year?
- Evergreen Gold has been rebranded as Evergreen//Forever. It's still the same Pure support and service that keeps your CapEx purchases current. Evergreen//Forever is the foundation for the next two offerings.
- Evergreen//One, previously known as Pure as-a-Service (PaaS), is Pure's OpEx consumption model that allows you to subscribe to storage for both on-prem and in the cloud storage needs. With as little as a 12-month commitment, you can get the storage capacity and performance you need to run your business without having to make a large up-front investment. Evergreen//One includes Evergreen//Forever, so when the storage controllers reach 3 years of use, they will be upgraded non-disruptively to the latest generation model of similar size.
While Evergreen//Forever and Evergreen//One seem to have you covered from a CapEx or OpEx point of view, there was a gap for the customers that fall in between and want both. Those businesses that either want (or need) to own the hardware but would like to subscribe to a consumption-based service offering for the software and services were left out.
- Enter Evergreen//Flex - To bridge the gap of CapEx and OpEx, so you can have your cake and eat it too. Consumers simply purchase (own) the hardware at a higher discounted rate to allow them to depreciate it via their normal depreciation schedules. They then can subscribe to the services they need for the capacity that's in use within their data centers. Evergreen//Flex also fixes the issue of stranded capacity in situations where you may have a storage array dedicated to a specific workload with more free capacity than needed. With Evergreen//Flex, you now have the ability to move the DirectFlash Modules (DFMs) around within your Fleet of Arrays while still staying covered with your consumption service contract.
While the following two weren't necessary new product announcements at techfest22, Pure Fusion and Portworx Data Services (PDS) went GA just prior to the yearly conference and are a big part of Pure's vision to support today's Modern Applications.
As today's applications have moved, or are moving to containers, yesterday's storage just doesn't keep up with the amount of changes occurring and at the speed the developers need those changes completed. This is where Portworx rises above the rest, by delivering a software-defined container native storage layer backed by local disks, storage arrays or cloud storage. Not only do your apps get the resiliency, replication, and performance from traditional storage arrays, but your apps can now move on/off premises and between cloud providers giving you the utmost in agility and freedom for choice on where you want your apps to run.
Portworx Data Services (PDS), the industry's first database-as-a-service platform. The goal for PDS is to simplify and shorten the time it takes to provision and maintain databases in your container environment. Your developers want to use the data services provided by the different databases, but they don't want to be the subject matter experts (SMEs) that supports them. PDS helps to remove the need for having the specialized skills to deploy and maintain databases like Cassandra, Couchbase, Elasticsearch, Kafka, Redis, RabbitMQ, MongoDB and PostgreSQL. Instead, these databases are delivered in a managed, as-a-service approach without locking you in and can be deployed via a self-service, on-demand management portal. With just a couple clicks you can deploy and/or update a database cluster following industry best practices without needing to know the specific complexities or requirements of the database.
But containerized applications are not the only applications needing a modern, self-service approach to accessing and managing their on-premises storage. Gone are the days of waiting for days to have storage provisioned manually for your application needs. Application owners want it now and don't want to wait.
Pure Fusion provides a cloud-like operating model for all your Pure Storage arrays by providing a Storage-as-Code™ experience for the app owners. By removing the need to manually monitor capacity usage and make storage placement decisions, Pure Fusion automates the volume creation and workload placement across your fleet of Pure Arrays while taking into considerations the storage tiers, location, and availability zones. The complete FlashArray product line, along with Cloud Block Store (CBS), are supported with Pure Fusion at launch along with support for multi-tenancy, automated workload re-balancing and SDKs for Python, Ansible and Terraform.
Environmental, Social and Governance (ESG) seems to be the buzz word these days across all verticals from both technology providers and technology consumers. Cities and countries are putting an emphasis on reducing the data center impacts on the environment with their power and cooling consumption and greenhouse gas emissions. With the cost of energy rising, it's front of mind for business owners to reduce the spend to run their business applications and store their data.
While most OEMs treat storage as a commodity, Pure Storage has a different view not only because it's better for the environment, but they believe that by pairing purpose-built software with custom designed hardware allows for your storage to get better over time. Pure Storage is leading the way with their ability to keep your storage current, which equates to the storage being in operation longer (97% of arrays deployed six years ago are still in service) and out of the e-waste landfills. By having the software and hardware tightly coupled, they are able to do more with less and deliver unbelievable performance and capacity from a small amount of rack space.
Pure's FlashArray line is seeing an over 75% reduction in energy and emissions compared to their competition's comparable product offerings. This includes the FlashArray//XL that was released in the end of 2021. Pure believes that the 5-year datacenter cost to support the storage array shouldn't cost more than the storage array itself, which unfortunately is true in some parts of the world where data center costs are through the roof for the competitions power-hungry storage arrays.
OK, but what's new with FlashBlade?
While many have said that the FlashBlade was ahead of its time and why would you want to store large unstructured file and object data on flash. Well, if you look at the market landscape now, the competition has been evolving their products to support commodity SSD drives along with front ending their spinning disks or SSD drives with SCM media to handle the metadata. The use of commodity equipment and expensive SCM media can only get you so far especially in the current world of supply chain issues were currently in. If you don't marry the hardware with the software or have Better Science*, your storage will get old quick and your ESG ratings will plummet.
The current FlashBlade is still alive and kicking with a scale-out cluster size supporting 7-150 blades. These blades were custom designed with storage on the chip, removing the need for flash translation layers and allowing the software to squeeze every bit of performance out of the hardware. As you grow, the performance and capacity scale linearly. However, one of the design changes that customers have been asking for is the ability to scale performance and capacity independently. As network, cpu and capacity continues to evolve and get faster and larger, so must the FlashBlade.
Introducing the FlashBlade//S (available now).
The "S" stands for //Speed, //Simplicity, //Scalability and //Sustainability all while delivering almost 2.5x the density, performance and power efficiency compared to other scale-out storage solutions on the market. The //S is a complete re-architecture of the hardware, while still using the same Purity//FB operating system. The new chassis design supports 10 Blades in 5 rack units (RU), but you may be asking yourself, didn't the original FlashBlade support 15 blades in 4 RU? And you would be correct. But considering that it would take 3 chassis, 36 blades and 12 RU for the original FlashBlade to equal what a single, fully populated chassis for the new FlashBlade//S can do for capacity and power, you quickly realize //Sustainability part of the new design.
Besides the chassis and the upgraded Fabric IO Modules (FIOMs), the blades themselves have been completely re-designed. As I mentioned prior, the blades in the original FlashBlade consisted of NAND storage soldered directly onto the blades and came in two sizes (17TB and 52TB) with the CPU cores sizes appropriately. With the modular disaggregated architecture (say that fives times fast) of the FlashBlade//S blades, the performance and capacity scale independently.
The S200 Blade is a general-purpose blade and will be the best option for about 80% of the existing FlashBlade workloads. The S500 blade is for your highest of performance needs and includes more CPU cores and RAM. Both blade models support 1-4 DirectFlash Modules (DFMs) in either 24TB or 48TB size. This equates to 40 DFMs across a 10-blade chassis or almost 2PB RAW in 5 RU at max config.
The FlashBlade//S uses the same QLC DFMs that are used today in the FlashArray//C and is able to deliver unbelievable performance without the need of using costly SCM media to front end the storage. This can only be done by utilizing DirectFlash Software with the DFMs to support the nearly infinite scale for their metadata architecture. As your needs grow, simply buy additional blades for more compute power or if capacity is what you are after, purchase additional DFMs to meet your needs. The FB//S is built for Exabyte scale while still delivering the best TB per RU density requiring less power and cooling per RU as well.
The FlashBlade//S along with NVIDIA's DGX Systems and NVIDIA's Networking make up the new AIRI//S – Pure and NVIDIA's partnership for a complete AI Ready Infrastructure solution. The proven reference architecture based design will allow you to start your deep learning initiatives quicker than trying to build-it yourself so your data scientists can remain productive and get their results faster.
Lastly, FlashBlade//S now supports Evergreen//Forever…. woohoo!! This means that the blade servers in the FlashBlade//S will be upgraded following the Ever Modern program mentioned above, the same way the controllers in the FlashArray are upgraded without downtime. While the original FlashBlade line does not support Evergreen//Forever, Pure still has you covered by providing an upgrade path via trade-in credits to move your investment in FlashBlade to the FlashBlade//S. In the end, this could be the last scale-out, file and object solution you'll ever need.
This one day Pure//Accelerate techfest22 conference packed a lot of punch in a short amount of time. With a bunch of exciting updates and announcements, it already has me looking forward to what's in store for next year. I believe that Pure's engineering philosophy and Evergreen architecture truly makes their solutions the only ones on the market that continually get better over time and never get old.
* Better Science: HW & SW Co-Design on-demand session from the Basecamp track.